HomeUpSearchMail
NEW

40m and 80m NVIS experiences

40m NVIS experiences by Victor Reijs is licensed under CC BY-NC-SA 4.0

Introduction

Definitions

SDR recievers in the neighborhood

Checked SDRs between 28/8/2026 15:15UTC and 7/9/2026 15:20UTC:

foF2 en fxI

Click below graph to get the latest version of the ionogram at Dourbes (Belgium).
Archives for the ionograms are here (click for URSI: DB049, select Year, Month and Day, and select the Measurement Time)

foF2 map

Here is an Earth ionogram map (from The Bureau of Meteorology) of the foF2 values; a pity there is no such map for fxI (one can approximate fxI by adding some 0.74MHz to foF2 (Male [2021] mentions 0.7MHz):

Measurement sessions

Each measurement session contained 17 data points:

• Date and time [UTC] (the time is rounded to the nearest Hour)
  
• TX Freq, on average 7.006MHz, depending on QRM, QRN or ongoing QSO
  
• of all five SDRs: signal and noise strength (in integer S-points)
• from Dourbes: foF2, fxI, hoF2(TX Freq) and hxI(TX Freq)

The TX was made ready by:

• checking if TX Freq is free (otherwise slightly change the TX Freq)
• send CW text: TEST PE1ATN/QRP NVIS PROP TEST N QSO QRL
  
• Tune a few seonds and record the signal strength and noise level (both in uncalibrated S-points) for each SDR
• send CW text: NVIS PROP TEST PE1ATN/QRP TINYURL.COM/36rkhh4b QRT
  

Signal strength and fxI related to the Sun's position

Signal strength (SS) [Witvliet, 2015,Fig 23]	fxI and foF2 [Witvliet, 2015,Fig 24]

G5RV junior behavior (40m-10m)

G5RV junior

Average height of G5RV junior as inverted V is around 7m (aka 0.175λ). The centre was at 9.3m and the ends are at 5m and 7m.
QTH PE1ATN: border of Lemelerveld.

• IMHO a more general approach (like non NVIS applications: e.g. resulting in an open spherical sector): Replace the sin(θ) with abs(sin(θ)) in Witvliet, formula (2) to (5) [2015]; use G(θ,ϕ) and D(θ,ϕ) from 0 to π; integrate θ over start to end of the cone angles; and integrate ϕ from 0 to π.
  More study how open sperical sectors are practical.
  
• 'D(θ,ϕ) is the directivity in the direction θ and ϕ, i.e., the antenna gain in that direction divided by the average antenna gain [using G(θ,ϕ) weighted by sin(θ)] over all possible spatial angles [of a sphere, as one compares with dBi], expressed as a linear value' [Witvliet, 2015, section 4.2].
  Text between brackets are additons given by Reijs.
• Resultaat van D_90,max formule (5) [Witvliet, 2015] (zie hierboven: orange line; 15.2dBi) lijkt groter te zijn dan die ik bereken met Simpson integratie en verschillende elevatie stap groottes bij gebruik van that the density per unit of solid angle (steradian) must be constant.
  

  Elevation step [deg]	Difference [dB]
  10	0.8
  5	0.4
  2.5	0.2
  1	0.1

  
  As G changes quite fast of a cone around 90deg; one needs elevation steps of 1deg to calculate D₉₀ properly (due to Simpson integration).
• The Far-Field (FF) diagrams for two different cones has been used: a cone that has equal power per sterradian (uniform solid sngle cone; orange curve: D_{90, max}) and an isotropic cut cone thas has equal power per angle (purple curve; D_{90,isotropic-cone}). See below picture:
  
  D_{90, max} is 15.2dBi and D_{90,isotropic-cone} is 13.5dBi.
  D_{nvis, max} (my D_{90, max}) is based around an uniform solid angle cone, which is the industry standard (IEEE Std 145-1993?).
  Personally I think that D_{90,isotropic-cone} is a handier reference, as it is close enough to most antenna behavior near 90deg (except for beamforming antenna). By the way: both values change when another cone is used (width and/or main direction).
  
• Reducing the elevation step (in Simpson integration) for 'normal' antenna FF-diagrams does not have much influence. Changing the step from 10deg to 1deg only changed G₉₀ with 0.01dB while the D₉₀ changed with 0.03dB. These changes will be small/neglectable for a reasonble flat FF-diagrams at NVIS angles.
  

Its SWR (measured with NanoVNA)

Difference NanoVNA and NEC-2 simulation

Measurement sessions

From 8/9/2026 to 9/9/2026 measurements were done every 5minutes to see why there are these fluctuations sometimes seen.

An initial analysis of the measurements

Date from 28/8/2026 until 2/9/2026:

:

NVIS dipole behavior (40m)

The design of this dipole came from here. Height of 40m dipole is around 1.45m (aka 0.035λ) and has two reflectors wires at ground level.
QTH PE1ATN/P: beside Overijssels kanaal near Lemelerveld

.

Its SWR (measured with NanoVNA)

Difference NanoVNA and NEC-2 simulation

Measurement sessions

See for the actual measurement sessions (up to now only 1) this page.

G5RV behavior (80m-10m)

Average height of G5RV as dipole is around 20m (aka 0.25λ).
QTH PE1ATN/P: at Scouting Lemelerveld.

EFHW behavior (80m-10m)

The lowest (start) point of EFHW is around 5m and the heighest (end) around 10m.
QTH PE1ATN: border of Lemelerveld.

So the EFHW has NVIS properties around 3.5 and 7MHz (simulated in EZNEC Pro2+ v.7.0; using NEC-2, elevation angle step=1deg).

Using Witvliet, formula (2) and (3) [2015], we get (elevation angle step=10deg):

Its SWR (measured with NanoVNA)

Difference NanoVNA and NEC-2 simulation

The SWR_s values look ok-ish. The f_rs is around +4% different for 80m.
<is being measured>

Measurement sessions

Findings

foF2 or fxI as threshold?

It looks that fxI is a better threshold value for possible communication than foF2. As soon as fxI drops below TX Freq., the propagation does down with some 25dB (from measurements of Reijs). Similar can be seen in Witvliet [2015, Fig.23 and 24]. The average height of the fxI layer is around 340km.
The fxI is on average 0.74MHz higher than foF2.

Max. effective elevation angle

The elevation is never very close to 90deg, as the underlaying layers (like E and F1) will have already refracted the wave front; so the wave front will not reach F2 at 90deg, but a few degrees less [Witvliet, 2015, section 3.1].

Sunrise/set relation with fxI and RX signal strength

After Sunise/set (at the location of the F-layer) one can see a slow change of the fxI and RX signal strength.

Signal strength comparison Witvliet en Reijs

The signal strength in Witvliet and Reijs are similar when including the extra TX of 22dB in Witfield; and both see a signal range-variation of some 15dB.
Both Witvliet en Reijs see that the signal strength in the evening is larger than in the morning: Witvliet some 11dB (Witvliet, 2015, Fig. 23] and Reijs some 8dB.

Propogation during night time

There is still propagation even when the fxI is much lower than the test frequency (like fxI-Freq = -2MHz). The night time signal strength of Witvliet is around 18dBμV (@850W) and for Reijs this is around 5dBμV (@5.5W). When including the TX strength diffence this becomes: 18dBμV (Witvliet) and 27dBμV (Reijs).
All waves transmitted during night time will be in some way scatter down and up and this will cause night-time scatter. This night-time scatter signal is much weaker than normal daytime NVIS: often 45 to 50 dB down from a true reflected. In the cause of Reijs and Witvliet the difference is respectivily some 25 and 35dB.

Propagation durig noontime

The dip in signal strength looks to be related to the fxI dip around noontime. The reason is the so-called noontime-bite-out (typically forced by thermospheric winds).The closer the fxI is to the TX Freq, the larger the signal strength dip looks to be.

Propogation early morning and early evening

The increase of the signal strength looks to happen around the Sunrise at 340km (location of the height of fxI layer). The decrease of the signal strength looks to happen around the Sunset at 340km (location of the height of fxI layer.
The duration of these increase and decrease are different (primarily due to a more rapid buildup of solar ultraviolet radiation in the morning) for Witvliet and Reijs (also knowing the difference in measurement accuracy (Witvliet Minutes; Reijs in Hours):

Signal strength fluctuations around when fxI comes close to TX Freq

It could be that fluctuations cause temporarily increase of signal strength (if fxI is close to TX Freq; at the time 17:25 on 8/9/2026 it shows clearly).
A similar effect (except the geometry [aka non-NVIS] is different) might have been seen with the frequency deviations in the Doppler measurements.
Witvliet [2015, Section 5.3] also sees such fluctuations of the fxI value: his 'hesitations'.

• Use SDR Maasbree (has a low noise number; sometimes close to S=1).
  
• 5sec before a 5min based time (as that is the interval of Dourbes ionogram takings) a CW tune until 5sec after that 5min based time.
• together with Dourbes data (in the middle of the CW tine) we can analyse its ionogram with the signal and noise numbers (in dB) from Maasbree.

Some measurements:

The Time axis looks continuous, but there are gaps in data points!

• Looking at the height of the F-layer, it varies in general between 300 and 440km and the extra attenuation of some 3.5dB/100km (using some 65 data points). This is not that far from the 2.5dB/100km as free space path loss.
• There looks to be some periodic peaking of fxI of between 70 and 110 minutes. This could cause these large fluctuations in signal strength around the time when fxI and TX Freq come close. This looks to be related to Traveling Ionospheric Disturbances (TIDs).

Effective NVIS antenna

Of the simulated 40m antennas (NVIS dipole, G5RV and EFHW); my G5RV junior looks to be the most effective (highest G₉₀ and D₉₀), followed by EFHW and then the NVIS dipole.

Simulation depending on calculating engine

EZNEC Pro2+ uses NEC-2 as the underlaying calculating engine. The ground-modelling is Sommerfeld-Norton.
The difference between Witvliet simulation [2015, Fig. 15] using NEC-4 and Reijs' simulation using NEC-2 is quite small (less than on average 0.2dB).
Would be interesting to see a better simulation for the f_rs when uisng NEC4 or NEC-5, as it can deviate some + or -10% at frequencies of 7MHz when using NEC-2.  If someone wants to help me running that, please let me know.

Calculating D_90,max

Code in Excel en R

In Excel a spreadsheet has been made which can receive the FF-diagram data and calculate the G₉₀ en D₉₀.
Based on that experience, a extended version has been made (with help of Google AI, although a lot of understanding and debugging is still needed!) with R that can: import EZNEC 3D FF data; import manual FF-diagram data; convert 3D FF-diagrams into an onmnidirectional FF-diagram; configurable elevation direction (elevation agle xx) and cone size (practicality of this still needs to be investigated); calculate G_xx en D_xx; and plot FF-diagrams with the calcuated info:

References

Chen, Yiding et al.: Latitudinal dependence of daytime electron density bite-out in the ionospheric F2-layer. In: JCR space physics (2020), issue 10.1029/2020JA028277, pp. 1-12.
Male, Jordi et al.: Analysis of the ordinary and extraordinary ionospheric modes for NVIS digital communications channels. In: Sensors 21  (2021), issue 2210.
Poole, Ian: Radio progation: Principles & practice. Herts, RSGB 2007.
Sapundjiev, Danislav and Stanimir Stankvc: Statistical analysis and modeling of the local ionospheric critical frequency: A mid-latitude single-station model for use in forecasting. In: Acta Geophysica 64 (2016)  issue 3, pp. 810-824.
Walden, M.C.: The extraordinary wave mode: Neglected in current practical literature for HF NVIS communications. In: Ionospheric radio Systems and Techniques.2009.
Witvliet, Ben A. et al.: Near vertical incidence skywave propagation: elevation angles and optimum antenna height for horizontal dipole antennas. In: IEEE Antennas and Propagation Magazine, 57  (2015), issue 1, pp. 1-18.

Acknowledgements

Disclaimer and Copyright
HomeUpSearchMail